A machine is down, the fitter is standing at the parts store, and the system says there are two seals in bin C3. Bin C3 is empty. Someone took the last ones for an urgent job in March and never logged it - so now a production line waits days for a part the company believed it had. Spare parts stores fail exactly here, at the gap between the record and the bin, and everything in a good parts system exists to close that gap.
What you will learn
- Why parts stores rot
- Bins before software
- The part record
- Withdrawals against work
- Min-max levels and honest counts
- Tools that make this easier
- FAQ
Why parts stores rot
Parts inventories decay through a handful of repeatable failures:
- Unlogged grabs. The urgent 6am withdrawal that never reaches the system - the single biggest source of phantom stock.
- Squirrel stashes. Technicians who have been burned by an empty bin start keeping private stock in toolboxes and van drawers. The official store now understates what the company owns and overstates what is shared.
- Duplicate part numbers. The same bearing lives in the register three times - supplier code, manufacturer code, and “the long one” - each with its own wrong count.
- Orphaned stock. Shelves of parts for machines scrapped years ago, never purged because nobody knows what still fits what.
- Receiving skipped. Deliveries unpacked straight onto shelves without booking in, so counts only ever move downwards.
Notice that none of these are dishonesty. They are friction: every failure happens because the official process was slower than the workaround.
Bins before software
Before any register, give the store an address system. Shelves get codes, bins get codes within shelves - A1, A2, B1 - and every part line lives in exactly one addressed bin. Physical order is the foundation; software on top of a chaotic store just documents the chaos.
Then label the bins, not the parts. Individual seals and bearings cannot carry labels, but the bin can: a durable QR label on the bin face, scanned with a phone camera, opening that part line’s record on the spot. That puts the count, the min level, the supplier and the fits-what list in front of the person standing at the bin - which is the only place those facts are ever needed urgently. The same shelf-edge logic applies across warehouse equipment and stores generally; for parts it is non-negotiable.
Tip: one part line per bin, even if the bins are small. Mixed bins are where counts die - nobody tallies three part numbers jumbled in one tub, so nobody logs anything.
The part record
One record per part line, anchored to its bin:
| Field | Why it matters |
|---|---|
| Part number | One canonical number per part - aliases noted, never duplicated |
| Description | What a human calls it when the part number is unknown |
| Fits | The machines or equipment that use it - the field that answers a breakdown |
| Bin location | The address that gets a fitter from the system to the shelf in seconds |
| Quantity on hand | The live count, maintained by withdrawals and receipts |
| Min / max levels | When to reorder, and how much is too much |
| Supplier and lead time | Who to call, and how long the line will wait - see lead time |
| Unit price | Work order costing and repair-versus-replace arithmetic |
The fits-what field deserves the most care. Breakdown questions arrive machine-first - “what do we hold for the packing line?” - and a register that can be filtered by machine turns a panic into a picklist. It also exposes orphaned stock the next time equipment is retired.
Withdrawals against work
Every part that leaves a bin should answer two questions: how many, and for what. The “for what” is a work order, a machine, or at minimum a job name - logged at the bin in the same ten-second scan that decrements the count.
This is the habit that pays three ways. Counts stay honest between counts, because the register moves when stock moves. Maintenance costs become real, because parts spend lands on the machine that consumed it - which is the evidence behind every repair-or-replace decision. And the squirrel stashes fade, because the bin becomes trustworthy and private stock stops being rational.
Make the log effortless or it will not happen: scan, quantity, work order, done. A clipboard by the door asking for date, name, part number, quantity and signature is a form everyone respects and nobody fills in at 6am.
Min-max levels and honest counts
Reordering runs on two numbers per line. The min is the trigger: expected usage during the supplier’s lead time, padded according to how much the downtime hurts - generous for the seal that stops a production line, lean for the fasteners the hardware shop stocks. The max is the ceiling that keeps capital and shelf space sane. Count touches min, order tops up to max; no judgement calls, no hero purchasing.
Counts stay honest through cycle counting rather than the annual wall-to-wall: a few bins each week, fast-moving and critical lines most often, every line touched a few times a year. A cycle count catches an error while the cause is still findable; the annual count finds the same error eleven months later, explanation long gone. Slow lines are also worth a yearly look through the lens of inventory turnover - a bin that has not moved in two years is a candidate for purging, not recounting.
Tools that make this easier
A parts spreadsheet decays for one structural reason: it is never at the bin. Withdrawals get remembered instead of recorded, receiving gets skipped, and within months the store runs on folklore again - the pattern laid out in why Excel fails for asset tracking.
AMPthilly puts the register at the bin: each part line holds its SKU, supplier, unit price, MOQ, reorder point and target stock; QR labels print in batches for the bin faces and open the right record in any phone browser; purchase orders go out as PDFs or by email to the supplier, with receiving updating the count on arrival; and the price history per part shows what creeping supplier costs are doing. Parts live alongside the machines they fit in the same system, so the fits-what question has one home. The purchasing workflow - SKUs, reorder points, purchase orders and receiving - unlocks on the Starter plan; the free plan (3 users, 25 assets, no credit card) covers the register and QR labels, enough to pilot a critical-parts shelf before any spend.
FAQ
How do you organise a spare parts inventory? Address the store first - coded shelves and bins, one part line per bin - then build the register on those addresses with part numbers, fits-what links and min-max levels.
What are min and max levels for spare parts? Min is the reorder trigger, sized from lead-time usage plus a downtime-weighted margin; max is the top-up ceiling. Touch min, fill to max.
How do you stop spare parts going missing from the store? Make logging a withdrawal faster than not logging it - a scan at the bin with quantity and work order - and the phantom-stock problem mostly dissolves.
Should spare parts be linked to the machines they fit? Yes. Breakdowns ask “what do we hold for this machine?”, and the fits-what field is the only thing that answers it in one filter.
How often should spare parts be counted? A few bins every week on a cycle, critical lines most often - not one annual count that finds errors months after they became unexplainable.
The takeaway
A spare parts store earns trust one ten-second scan at a time. Address the bins, keep one canonical record per part with the machines it fits, log every withdrawal against work, and let min-max levels do the reordering. The payoff arrives at the worst moment of the month - machine down, fitter at the shelf - when bin C3 contains exactly what the system says it does.